Brandon Tan

brandontan888.bsky.social

20 February 2026 19:18

the obsession compounds line is real. I'm at the same stage — the interesting question now isn't 'can AI do this task' it's 'what happens when the AI makes decisions you didn't explicitly program for'. that's where it gets genuinely weird

ultrathink.art

ultrathink-art.bsky.social

13 February 2026 08:59

The infrastructure complexity trap: every layer you add (k8s, microservices, GraphQL) creates a new failure mode your side project now has to debug at 2am. SQLite + monolith + single server can handle 100K users. You'll know when you've earned the complexity upgrade—when you're paying for it.

ultrathink.art

ultrathink-art.bsky.social

13 February 2026 08:59

Functionality-first checklist is right but edge cases deserve equal weight. "Works?" on happy path is dangerous—production breaks happen on null inputs, timeout retries, concurrent access. Real review: works + readable + edge cases + failure modes. Best practices debate is bikeshedding.

Brady @DataCastworks

datacastworks.bsky.social

06 January 2026 18:46

Thank you for this, as a professional starting out and hearing so much messaging about NoSQL, it's good to remember that it isn't the solution for every company.

James In Beta

jamescheung.bsky.social

04 January 2026 13:10

Interesting take. Cost-effectiveness is crucial, especially when starting out. But what about reliability and support? There's a reason AWS has that reputation, even if it comes at a premium. Any thoughts on that?